Explore_data_for_social information

pacman::p_load(tidyverse, rjson, data.table, gghalves, plotly, gganimate, av, colorspace)
`%!in%` = Negate(`%in%`)

Identify participants that can provide different types of social information

relevant variables

• n of switches
• n of consecutive clicks
• n of gems founds

n of point searned vs. expected points (better than chance)

finding social info:

criteria: - points are representative to average - gem not found before round 10 - exploration is conistent

find average points when a gem is found

points_when_gem_found <- explore_data %>% 
  group_by(player, gempresent, tot_points, env_number) %>% 
  filter(gempresent == 1 & sum(gem) > 1) %>% 
  select(player, gempresent, tot_points, unique_rounds, env_number) %>% 
  distinct() %>% 
  ungroup() %>% 
  group_by(env_number) %>% 
  mutate(performance_group = ntile(tot_points, 3))


# distribution of point when gem is found

points_when_gem_found %>% 
  ggplot() +
  geom_histogram(aes( x = tot_points, fill = factor(performance_group)), binwidth = 40)+
 #, color = "white") +
  #geom_vline(aes(xintercept = mean(tot_points)), lty = 2, size = 1, color = 'black') + 
  #geom_half_point(aes(x = factor(gempresent), y = points), alpha = 0.02) +
  #theme_bw(base_size = 20) +
  facet_wrap(~env_number)

  guides(fill = "none")

## $fill
## [1] "none"
## 
## attr(,"class")
## [1] "guides"

timing

now filter for gem not found before round 10

# list of players 

avg_rounds <- points_when_gem_found %>% 
  filter(performance_group == 2 | performance_group == 1) %>% 
  ungroup() %>% 
  select(unique_rounds)

avg_rounds <- avg_rounds$unique_rounds

early_gems_idx <- which(explore_data$gem & explore_data$trial <= 5)
round_to_exclude <- unique(explore_data$unique_rounds[early_gems_idx])


late_gems <- explore_data %>% 
  filter(unique_rounds %in% avg_rounds)%>% 
  filter(unique_rounds %!in% round_to_exclude)


how_may_late_gems <- late_gems%>% 
  group_by(env_number) %>% 
  summarise(sum = n())

visualize social info

make this into shiny app

gems_coords <- explore_data %>% 
  filter( points >160) %>% 
  select(x , y, env_number, round) %>% 
  filter(player == 42 & round == 2) %>%
  distinct()

## Adding missing grouping variables: `player`

one_sequence <- late_gems %>%
  filter(player == 42 & round == 2) %>%
  ggplot() +
  #geom_bin_2d(aes(x = x, y = y, fill = gem), binwidth = c(1, 1)) +
    geom_rect(
    data = gems_coords,
    size = 1,
    colour = "red",
    fill = "white",
  
    aes(
      xmin = x - 1,
      xmax = x  ,
      ymin = y - 1,
      ymax = y
    )
  ) +
    geom_rect(aes(
    xmin = x - 1,
    xmax = x  ,
    ymin = y - 1,
    ymax = y,
    fill = points,
    frame = trial
  ),
  color = "black",
  size = .2) +
  
  scale_x_continuous(limits = c(-1, 7), breaks = -1:7) +
  scale_y_continuous(limits = c(-1, 7), breaks = -1:7) +
  scale_fill_continuous_divergingx(palette = "RdBu", mid = 0)

## Warning: Ignoring unknown aesthetics: frame

  ggplotly(one_sequence) %>% 
          animation_opts( transition = 0)